Skip to main content
Theorem
Path
Curriculum
Paths
Labs
Diagnostic
Case Study
Blog
Search
Sign in
Quiz Hub
/
Bias-Complexity Tradeoff
Bias-Complexity Tradeoff
5 selected
Difficulty 3-5
5 unseen
View topic
Foundation
New
0 answered
1 foundation
4 intermediate
Adapts to your performance
Question 1 of 5
120s
foundation (3/10)
state theorem
In the SSBD Section 5.2 decomposition
L
D
(
h
S
)
=
ε
app
+
ε
est
, what is the *approximation error*
ε
app
?
Hide and think first
A.
The Bayes risk
L
D
Bayes
, the irreducible noise floor under any function from
X
to
Y
for the loss.
B.
The minimum population risk over the class,
min
h
∈
H
L
D
(
h
)
, depending on
H
and
D
but not on
m
.
C.
The supremum population-vs-empirical-risk gap
sup
h
∈
H
∣
L
D
(
h
)
−
L
S
(
h
)
∣
on the training sample.
D.
The expected gap between the ERM and best-in-class hypothesis on a fresh sample, taken over the random training sample of size
m
.
Show Hint
Submit Answer
I don't know